Central Catadioptric Camera Calibration using Planar Objects

Abstract

Central catadioptric cameras combine lenses with mirrors to enlarge the field of view while keeping a single effective viewpoint. In this paper we propose a novel method of calibrating the intrinsic parameters of central catadioptric cameras using a planar object. Based on the viewing sphere model, we can warp a portion of the catadioptric image to an image captured by a virtual perspective camera with given intrinsic and extrinsic parameters. We show that placing the planar object several times around the catadioptric camera is equivalent to placing the same object at different poses relative to a static virtual perspective camera. Therefore, homography method can be applied to calculate the relative poses of the planar object as well as the projection error of feature points on the planar object. By minimizing the projection error, we can obtain the optimized intrinsic parameters of the central catadioptric camera. Besides simplicity, experiments with simulation and real image data clearly demonstrate the high robustness and accuracy of the proposed calibration method.